Hydrographic biases in global coupled climate models and their relation to the meridional overturning circulation

Comparison of the volumetric theta/S distribution of models participating in the Climate Model Intercomparison Project 3 (CMIP3) indicates that these models differ widely in their ability to represent the thermohaline properties of water masses. Relationships between features of the quasi-equilibrium hydrographic mean state of these models and aspects of their overturning circulations are investigated. This is achieved quantitatively with the help of seven diagnostic hydrographic stations. These few stations were specifically selected to provide a minimalist schematic of the global water mass system. Relationships between hydrographic conditions in the North Atlantic measured with a subset of these stations suggest that hydrographic properties in the subpolar North Atlantic are set by the circulation field of each model, pointing towards deficiencies in the models ability to resolve the Gulf Stream-North Atlantic Current system as a major limitation. Since diapycnal mixing and viscosity parameterizations differ across CMIP3 models and exert a strong control on the overturning, it is likely that these architectural differences ultimately explain the main across-model differences in overturning circulation, temperature and salinity in the North Atlantic. The analysis of properties across the quasi-equilibrium states of the CMIP3 models agrees with previously reported relationships between meridional steric height gradients or horizontal density contrasts at depth and the strength of the deep water cell. Robust relationships are also found in the Southern Ocean linking measures of vertical stratification with the strength of the abyssal circulations across the CMIP3 models. Consistent correlations between aspects of the quasi-equilibrium hydrography in the Southern Ocean and the sensitivity of the abyssal cell to increasing radiative forcing by 2100 were found. Using these relations in conjunction with modern hydrographic observations to interpolate the fate of the abyssal cell suggests that the Southern abyssal cell may decrease by roughly 20 % by the end of the century. Similar systematic relationships between the quasi-equilibrium hydrographic states of the models and the sensitivity of their Atlantic deep water cell could not be found.